2025 Volume 89 Issue 4 Pages 457-462
Background: Frailty is highly prevalent and associated with a poor prognosis in older patients with heart failure (HF). In this study, we investigated the association between frailty, as assessed by the Kihon Checklist (KCL), and readmissions in older patients with HF.
Methods and Results: We performed a retrospective cohort study of all consecutive older patients hospitalized for HF aged ≥65 years between September 2016 and March 2018. The KCL was based on the health condition and living situation of each patient prior to hospitalization and was categorized into 4 groups based on quartiles of the total score (Q1–4). The primary outcome was readmission due to HF within 2 years post-discharge. A total of 244 patients (111 males; mean age, 81.7 years [6.9]) were included. During 2 years of follow-up post-discharge, 71 patients (29.1%) experienced an adjudicated readmission for acute HF. Multivariable Cox regression analysis revealed that Q2–4 of the KCL were associated with an increased hazard ratio (HR) for HF readmission when compared with Q1 (Q2; HR [95% confidence interval (CI)]: 9.54 [2.78–32.66], P<0.001; Q3; 8.28 [2.37–28.84], P<0.001; Q4; 9.12 [2.51–33.11], P<0.001).
Conclusions: Our findings revealed an association between frailty, as assessed by the KCL, and readmissions for HF within 2 years of discharge in older patients with HF.
An estimated 64 million people worldwide suffer from heart failure (HF), thus representing a major health concern.1 Moreover, although the per capita rate for hospitalized HF has begun to decline in the USA and several European countries, early post-discharge mortality and readmission rates remain largely unchanged, and may even be worsening.2 Frailty is one of several risk factors for a higher risk of HF readmission and death.3 A previous meta-analysis of 26 studies, featuring a total of 6,896 patients, estimated the prevalence of frailty in patients with HF to be 44.5% [95% confidence interval (CI), 36.2–52.8%].4 Therefore, the importance of evaluating frailty in patients with HF has become clearly evident due to its high prevalence and prognostic effect. Moreover, a large observational study (FRAGILE-HF), featuring 15 centers in Japan, reported that almost all patients with HF experienced problems in multiple physical, cognitive, and social domains; a greater number of frailty domains was associated with a higher mortality rate and greater risk of HF readmission.5 Therefore, frail patients with HF require a comprehensive clinical approach that does not simply focus on disease-specific management.6
At present, several approaches are used to evaluate frailty; the 2 most common approaches are the phenotype model and the accumulation deficit model. The phenotype model is based on a predefined set of 5 criteria that are used to evaluate the presence or absence of signs or symptoms.7 The model has been applied in many studies of frailty in patients with heart disease, including the FRAGILE-HF study.5,8,9 In contrast, the accumulation deficit model considers the gradation of frailty with the progressive accumulation of deficits, each of which is given an equal weighting. The accumulation deficit model enables consideration of frailty in terms of stages rather than simply presence or absence. Furthermore, when applying this model, adverse outcomes are related to the number of equally weighted deficits as a measure of accumulated frailty, rather than specific clusters of deficits.10
Recently, the Kihon Checklist (KCL) has been used as a specific assessment tool for frailty, and its predictive validity for evaluating long-term care and death in community-dwelling older adults has been reported.11 The KCL was developed by the Ministry of Health, Labour and Welfare in Japan to identify older adults at high risk of requiring long-term care. The KCL features 25 yes/no questions in 7 areas: activities of daily living, mobility, nutrition, oral function, confinement, cognitive function, and depression. The underlying principle of the KCL is similar to the concept of the accumulation deficit model, which assesses frailty progression as a continuous variable, in that it evaluates frailty as the accumulation of geriatric syndromes based on a total score.12 The accumulation of geriatric syndromes incorporated in the KCL, including oral function, confinement, and depression, collectively constitute frailty in older patients and may increase the risk of HF readmissions. However, previous studies of frailty in patients with HF have mostly applied the phenotype model8 and did not consider oral function, confinement or depression, even when applying the accumulation deficit model.13 The specific relationship between frailty, as assessed by the KCL, and HF readmission has yet to be fully elucidated.
The purpose of this study was to investigate the association between frailty status, as assessed by the KCL, and readmissions in older patients with HF by applying a retrospective cohort design. We hypothesized that patients with HF who had a higher number of applicable items on the KCL, indicating progression of frailty, would have a higher risk of readmission.
We performed a retrospective cohort study of all consecutive older patients who were hospitalized for HF between September 2016 and March 2018. Clinical data were collected from medical records, including information relating to the etiology of HF, comorbidities, and the use of medications. Baseline physical findings (including body mass index [BMI]), blood samples, and echocardiography were obtained at discharge.
Study PopulationAll patients were evaluated for eligibility according to specific inclusion criteria: aged ≥65 years and able to walk alone without assistance from another person (including those able to walk with a cane or other walking aid). The exclusion criteria were: a history of heart transplantation or the current use of a left ventricular assist device; severe cognitive impairment, defined as a Mini-Mental State Examination (MMSE) score <17;14 and difficulty in responding to questionnaires. All patients underwent comprehensive cardiac rehabilitation as part of their disease management program. Due to the opt-out option provided on the institution’s home webpage, the requirement for additional informed consent to participate in this study was deemed unnecessary. This study was conducted in accordance with the Declaration of Helsinki and the Japanese Ethical Guidelines for Life Sciences and Medical Research Involving Human Subjects. The study protocol was approved by the ethics committees of Kitano Hospital (Approval No: 2110010) and Osaka Metropolitan University (Approval No: 2023–104).
Exposure: FrailtyFrailty was assessed by the KCL.11 Each item is assigned a value of 1 (yes) or 0 (no), resulting in a total score ranging from 0 to 25. A higher score indicates a greater degree of worsening frailty. Although the KCL includes a smaller number of items than the standard Frailty Index (FI), it has been demonstrated to be highly correlated with the FI.15 In this study, the KCL was completed based on the health condition and living status of each patient prior to hospitalization and was evaluated when HF had stabilized following initial treatment.
Primary Outcome: HF ReadmissionWe retrospectively collated the prognoses of all registered patients, within 2 years of discharge. The prognostic primary outcome was HF readmission. Following discharge, most patients were followed up in outpatient clinics at least every 3 months. Follow-up was based on medical need. For those without hospital follow-up, prognostic data were obtained from the medical records of other medical facilities that cared for the patient. Censoring due to dropout was defined as the last visit to a hospital, regardless of the dropout or the reason for the visit.
Potential CovariatesPotential confounders were identified by a directed acyclic graph based on expert knowledge and literature. The variables included age, sex, history of heart failure (presence or absence), living arrangement (living with others or alone), diabetes (presence or absence), New York Heart Association (NYHA) functional classification I/II or III/IV, left ventricular ejection fraction (LVEF), B-type natriuretic peptide (BNP), estimated glomerular filtration rate (eGFR), and MMSE score at discharge. BNP level ≥200 pg/mL was defined as elevated.
Statistical AnalysisBaseline characteristics are described by means and standard deviations or, in the case of categorical data, numerical values and percentages. To investigate dose-response associations, we categorized the KCL score into quartiles (Q) of the total score (Q1: 0–7 points; Q2: 8–11 points; Q3: 12–14 points; Q4: 15–22 points). Kaplan-Meier estimation was used to plot the cumulative incidence of HF readmission across the KCL score quartiles. A log-rank test was used for group comparisons. For time-to-event analyses, we conducted cause-specific Cox proportional hazards regression analysis to account for death as a competing risk and to investigate the specific association between HF readmission and KCL score (categorical variable). To properly account for death as a competing risk, data for non-survivors were censored at a time point beyond that of the last surviving patient.16 The proportional hazard assumption was evaluated by generating a log-log plot. We also calculated hazard ratios (HRs) and 95% CIs. Cox regression analysis was adjusted to account for potential confounding variables and used to estimate the HRs for HF readmission. In order to assess the robustness of our study, we performed 2 sensitivity analyses. First, we performed Cox regression analysis to investigate the association between the combined endpoint of HF readmission, all-cause death, and the KCL score. Second, we repeated the Cox regression analysis but excluded patients who were readmitted for HF during the initial month of follow-up; this practice minimized reverse causation bias. All statistical analyses were performed using SPSS version 21.0 software (IBM Corporation, Tokyo, Japan), and EZR software version 1.61 (Saitama Medical Center, Jichi Medical University, Saitama, Japan).
A total of 284 patients with HF were potentially eligible for inclusion, but after 39 patients were excluded, there were 245 patients who met the eligibility criteria. One of them died during hospitalization, resulting in a final study cohort of 244 patients for analysis (111 men, 133 women; mean [SD] age, 81.7 [6.9] years, mean [SD] LVEF, 48.3 [18.7]) (Figure 1). The baseline characteristics of patients, according to KCL quartile, are given in Table 1. Patients with higher KCL scores were older and associated with an increased proportion of NYHA class III/IV. The MMSE score declined as the KCL score increased. With regards to biomarkers, a higher KCL score was associated with a lower eGFR. However, BMI, LVEF, the proportion of underlying cardiac disorders, DM, and BNP, did not differ significantly when compared between quartile groups categorized by KCL score. The most prevalent underlying cardiac disorder was arrhythmia.
Flowchart of patients’ eligibility. MMSE, Mini-Mental State Examination.
Baseline Characteristics of Patients by Quartiles of KCL Score
| Characteristic | Quartiles of the KCL score | ||||
|---|---|---|---|---|---|
| Total (n=244) |
Quartile 1 (n=53) |
Quartile 2 (n=64) |
Quartile 3 (n=52) |
Quartile 4 (n=75) |
|
| Age (years) | 81.7 (6.9) | 78.3 (6.6) | 81.9 (6.8) | 82.8 (5.7) | 83.1 (7.0) |
| Female sex, n (%) | 133 (54.5) | 24 (45.2) | 32 (50.0) | 29 (55.7) | 48 (64.0) |
| Kihon Checklist score, median (IQR) | 12 (0–22) | 5 (0–7) | 9 (8–11) | 13 (12–14) | 16 (15–22) |
| Past HF hospitalization, n (%) | 83 (34.0) | 13 (24.5) | 20 (31.2) | 20 (38.4) | 30 (40.0) |
| Living alone, n (%) | 74 (30.3) | 17 (32.0) | 20 (31.2) | 19 (36.5) | 18 (24.0) |
| NYHA class III/IV, n (%) | 125 (51.2) | 9 (16.9) | 36 (56.2) | 28 (53.8) | 57 (76.0) |
| BMI (kg/m2) | 21.5 (3.8) | 21.7 (3.4) | 21.3 (4.0) | 21.9 (4.0) | 21.1 (3.8) |
| Left ventricular ejection fraction (%) | 48.3 (14.7) | 46.7 (15.7) | 49.6 (14.9) | 48.3 (13.4) | 48.1 (14.8) |
| MMSE (points) | 25.8 (3.9) | 27.9 (2.0) | 26.1 (3.1) | 26.3 (3.2) | 23.4 (3.9) |
| Etiology | |||||
| Ischemic heart disease, n (%) | 58 (23.7) | 13 (24.5) | 16 (25.0) | 14 (26.9) | 15 (20.0) |
| Hypertensive heart disease, n (%) | 16 (6.5) | 3 (5.6) | 3 (4.6) | 3 (5.7) | 7 (9.3) |
| Valve heart disease, n (%) | 12 (4.9) | 10 (18.8) | 14 (21.8) | 18 (34.6) | 17 (22.6) |
| Arrhythmia, n (%) | 73 (29.9) | 17 (32.0) | 21 (32.8) | 11 (21.1) | 24 (32.0) |
| DCM/HCM, n (%) | 12 (4.9) | 2 (3.7) | 3 (4.6) | 2 (3.8) | 5 (6.6) |
| Comorbidity | |||||
| Hypertension, n (%) | 133 (54.5) | 31 (58.4) | 35 (54.6) | 28 (53.8) | 39 (52.0) |
| Diabetes mellitus, n (%) | 71 (29.0) | 19 (35.8) | 13 (20.3) | 15 (28.8) | 24 (32.0) |
| COPD, n (%) | 19 (7.7) | 3 (5.6) | 4 (6.2) | 8 (15.3) | 4 (5.3) |
| Cancer, n (%) | 21 (8.6) | 6 (11.3) | 6 (9.3) | 3 (5.7) | 6 (8.0) |
| Prescription at discharge | |||||
| β-blocker, n (%) | 175 (71.7) | 33 (73.5) | 45 (70.3) | 32 (61.5) | 59 (78.6) |
| ACE-i/ARB, n (%) | 107 (43.8) | 25 (39.2) | 26 (40.6) | 20 (38.4) | 29 (38.6) |
| MRA, n (%) | 86 (35.2) | 23 (43.3) | 22 (34.3) | 20 (38.4) | 21 (28.0) |
| Loop diuretic, n (%) | 221 (90.5) | 40 (92.4) | 58 (90.6) | 48 (92.3) | 66 (88.0) |
| Statin, n (%) | 88 (36.0) | 19 (35.8) | 24 (37.5) | 22 (42.3) | 23 (30.6) |
| Laboratory data at discharge | |||||
| Hemoglobin (g/dL) | 11.3 (1.9) | 11.8 (2.2) | 11.4 (1.8) | 11.4 (1.9) | 10.9 (1.8) |
| Albumin (g/dL) | 3.7 (0.3) | 3.8 (0.4) | 3.7 (0.3) | 3.6 (0.4) | 3.6 (0.3) |
| Creatinine (mg/dL) | 1.5 (1.4) | 1.6 (2.1) | 1.2 (0.7) | 1.3 (0.9) | 1.7 (1.6) |
| eGFR (mL/min/1.73 m2) | 41.6 (18.6) | 46.0 (19.0) | 43.4 (16.0) | 40.9 (18.9) | 37.2 (19.8) |
| BNP (pg/mL) | 233.0 (189.8) | 227.0 (230.8) | 211.3 (145.9) | 203.7 (160.6) | 275.8 (205.2) |
Data are presented as mean (SD) unless stated otherwise. ACE-i, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; BNP, B-type natriuretic peptide; COPD, chronic obstructive pulmonary disease; DCM, dilated cardiomyopathy; eGFR, estimated glomerular filtration rate; HCM, hypertrophic cardiomyopathy; KCL, Kihon Checklist; MMSE, Mini-Mental State Examination; MRA, mineralocorticoid receptor antagonist; NYHA, New York Heart Association.
Association Between KCL Score and the Primary Outcome
The mean follow-up period was 490 days, during which 71 patients (29.1%) experienced an adjudicated readmission for acute HF, and 24 patients (9.8%) died (adjudicated all-cause death). In total, 30 patients were censored during the follow-up phase due to dropout. Figure 2 shows the Kaplan-Meier curves for different KCL scores. The rates of HF readmission after 2 years were 0.032 per year for patients in Q1, 0.293 per year in Q2, 0.265 per year for Q3, and 0.302 per year for Q4. Multivariable-adjusted cause-specific Cox regression analysis revealed that quartiles 2–4 of the KCL were associated with an increased hazard ratio for HF readmission when compared with Q1 after adjusting for potential confounding variables (Q2; HR [95% CI]: 9.54 [2.78–32.66], P<0.001; Q3; 8.28 [2.37–28.84], P<0.001; Q4; 9.12 [2.51–33.11], P<0.001) (Table 2). Sensitivity analyses revealed that patients with a high KCL score (≥8 points) had a significantly increased HR of the composite endpoint (Q2; HR [95% CI]: 5.87 [2.34–14.72], P<0.001; Q3; 4.92 [1.92–12.60], P<0.001; Q4; 5.41 [2.01–14.55], P<0.001) (Table 3). Finally, we performed sensitivity analysis to address potential reverse causation bias by excluding patients who were readmitted for HF within the first month of follow-up. These analyses yielded comparable conclusions (Q2; HR [95% CI]: 8.98 [2.62–30.08], P<0.001; Q3; 7.32 [2.08–25.75], P=0.001; Q4; 8.65 [2.37–31.57], P=0.001) (Table 4).
Kaplan-Meier curves for heart failure readmission for patients in different Kihon Checklist score quartiles.
HRs and 95% CIs for HF Readmission by KCL Score
| Variable | Patients (n) |
Frequency of HF readmission |
Occurrence rate/year |
Unadjusted Cox model | Adjusted Cox model | ||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | ||||
| Quartile 1 | 53 | 3 | 0.032 | 1 (Ref.) | 1 (Ref.) | ||||
| Quartile 2 | 64 | 24 | 0.293 | 4.91 | 2.48–27.44 | 0.001 | 9.54 | 2.78–32.66 | <0.001 |
| Quartile 3 | 52 | 19 | 0.265 | 4.52 | 2.28–26.14 | 0.001 | 8.28 | 2.37–28.84 | <0.001 |
| Quartile 4 | 75 | 25 | 0.302 | 4.81 | 2.52–27.77 | 0.001 | 9.12 | 2.51–33.11 | <0.001 |
Adjusted variables for HF readmission were as follows: age; sex; living alone; left ventricular ejection fraction; history of HF; diabetes; B-type natriuretic peptide; estimated glomerular filtration rate; Mini-Mental State Examination; and New York Heart Association class III/IV at discharge. CI, confidence interval; HF, heart failure; HR, hazard ratio; KCL, Kihon Checklist.
HRs and 95% CIs for Combined Events by KCL Score
| Variable | Unadjusted Cox model | Adjusted Cox model | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| Quartile 1 | 1 (Ref.) | 1 (Ref.) | ||||
| Quartile 2 | 4.91 | 2.03–11.88 | <0.001 | 5.87 | 2.34–14.72 | <0.001 |
| Quartile 3 | 4.52 | 1.83–11.16 | 0.001 | 4.92 | 1.92–12.60 | <0.001 |
| Quartile 4 | 4.81 | 1.99–11.63 | <0.001 | 5.41 | 2.01–14.55 | <0.001 |
Adjusted variables for the combined event were as follows: age; sex; living alone; left ventricular ejection fraction; a history of HF; diabetes; B-type natriuretic peptide; estimated glomerular filtration rate; Mini-Mental State Examination; and New York Heart Association class III/IV at discharge. Abbreviations as in Table 2.
HRs and 95% CIs for HF Readmission by KCL Score, Excluding Patients Who Were Readmitted for HF Within the First Month of Follow-up
| Variable | Unadjusted Cox model | Adjusted Cox model | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| Quartile 1 | 1 (Ref.) | 1 (Ref.) | ||||
| Quartile 2 | 8.03 | 2.41–26.78 | 0.001 | 8.98 | 2.62–30.08 | <0.001 |
| Quartile 3 | 7 | 2.05–23.91 | <0.001 | 7.32 | 2.08–25.75 | 0.001 |
| Quartile 4 | 7.89 | 2.36–26.33 | 0.001 | 8.65 | 2.37–31.57 | 0.001 |
Adjusted variables for the HF readmission were as follows: age; sex; living alone; left ventricular ejection fraction; a history of HF; diabetes; B-type natriuretic peptide; estimated glomerular filtration rate; Mini-Mental State Examination; and New York Heart Association class III/IV at discharge. Abbreviations as in Table 2.
In this study, we used a retrospective cohort design to investigate the specific association between frailty status, as assessed by the KCL, and readmissions in patients with HF within 2 years of discharge. Our analyses found that patients in the second quartile (or higher) of the KCL (KCL score ≥8 points) were associated with a HR for HF readmission that was approximately 8–9-fold higher than that for patients in the first quartile (KCL score <8 points). In addition, we conducted several sensitivity analyses, including analyses that accounted for death as a competing risk or minimized reverse causation bias; the outcomes of these analyses supported our primary findings. These results are consistent with those published previously by the FRAGILE study,5 which reported that the greater the number of frailty domains, the higher the mortality rate and HF readmissions. Taken together, the findings highlight the importance of implementing a more comprehensive assessment of frailty in older patients with HF.
The findings of this study also indicate that the accumulation of geriatric syndromes, as assessed by the KCL, increases the HR of HF readmission; this is a novel finding that highlights the importance of evaluating frailty in patients with HF from the perspective of accumulated deficit models. Although many previous studies have investigated the relationship between frailty based on the phenotype model and HF readmission,17 there is growing interest in evaluating and treating frailty from a more comprehensive perspective. The concept of the KCL is similar to that of the accumulation deficit model in terms of assessing the degree of frailty progression, not only in terms of mobility and cognitive functions, but also in terms of the accumulation of geriatric syndromes, including impaired activities of daily living, low nutrition, poor oral function, confinement, and depression. The accumulation of various geriatric syndromes has been shown to reduce the capability of a patient to respond to new stressors, thus leading to the onset and progression of chronic diseases.18 In addition, Rockwood et al. proposed that the accumulation of equally weighted deficiencies is the key factor associated with adverse outcomes, rather than a specific deficiency.18,19 A longitudinal study of community-dwelling older adults previously reported that older adults with multiple geriatric syndromes, as assessed by the KCL, had a shorter healthy life expectancy.20 Our current findings suggest that the frailty of patients with HF should be assessed as an accumulation of geriatric syndromes in community life, as with the KCL evaluation items, and not simply based on the presence or absence of signs (the phenotype model)17 or the accumulation of disease-related deficits (e.g., medical history and blood samples).13
Another noteworthy aspect to be considered is that the HR for HF readmission was comparable between patients in the second, third, and fourth quartiles of the KCL score. Similar to our findings, previous studies in patients with chronic diseases also identified a nonlinear relationship between the FI and outcomes, with the risk plateauing after a certain threshold (cutoff point: 0.25–0.3), even as the FI increased.21 Prior studies have also used a cutoff of 8 points to determine frailty in the KCL,11 which is consistent with the Q1 and Q2–4 thresholds in this study. In other words, a KCL score ≥8 points may indicate a level of vulnerability in which the accumulation of deficits exceeds a specific threshold and will lead to HF readmission.
Study LimitationsFirst, it must be acknowledged that the survey responses to the KCL are subject to a degree of recall bias. Second, KCL is a self-reported assessment and does not necessarily reflect actual conditions. To address these issues and mitigate the potential for misclassification, future prospective cohort studies should focus specifically on objectively evaluating patient frailty prior to hospital admission.
Our analysis identified a longitudinal association between frailty status, as assessed by the KCL, and HF readmissions within 2 years of discharge for older patients with HF. Individuals with a KCL score ≥8 points were associated with progression of frailty via the accumulation of geriatric syndromes and exhibited an increased HR for HF readmission. Our findings highlight the importance of evaluating frailty to prevent the readmission of patients with HF, not only from the perspectives of physical function or disease-specific deficits, but also as an accumulation of geriatric syndromes in multiple domains, thus allowing us to provide a more comprehensive level of care.
We are grateful to the patients for their cooperation in the study. We thank the collaborating investigators for their contribution.
There are no conflicts of interest to declare.
Kitano Hospital (Approval No: 2110010) and Osaka Metropolitan University (Approval No: 2023–104).
A single-center study at the Kitano Hospital in Osaka, Japan.
